NetBSD Problem Report #39642
From root@azeotrope.org Sun Sep 28 02:10:38 2008
Return-Path: <root@azeotrope.org>
Received: from mail.netbsd.org (mail.netbsd.org [204.152.190.11])
by narn.NetBSD.org (Postfix) with ESMTP id 72D2A63B884
for <gnats-bugs@gnats.NetBSD.org>; Sun, 28 Sep 2008 02:10:38 +0000 (UTC)
Message-Id: <20080928021033.2218D14DE@sloth.azeotrope.org>
Date: Sat, 27 Sep 2008 21:10:32 -0500 (CDT)
From: khym@sloth.azeotrope.org
Reply-To: khym@sloth.azeotrope.org
To: gnats-bugs@gnats.NetBSD.org
Subject: tcom(4) driver is broken
X-Send-Pr-Version: 3.95
>Number: 39642
>Category: kern
>Synopsis: tcom(4) driver is broken
>Confidential: no
>Severity: serious
>Priority: medium
>Responsible: kern-bug-people
>State: closed
>Class: sw-bug
>Submitter-Id: net
>Arrival-Date: Sun Sep 28 02:15:00 +0000 2008
>Closed-Date: Sun Sep 28 15:41:08 +0000 2008
>Last-Modified: Sun Sep 28 19:50:02 +0000 2008
>Originator: Dave Huang
>Release: NetBSD 4.99.72
>Organization:
>Environment:
System: NetBSD sloth.azeotrope.org 4.99.72 NetBSD 4.99.72 (FOXY) #4: Sat Sep 27 20:02:31 CDT 2008 khym@cheetah.azeotrope.org:/usr/obj.i386/sys/arch/i386/compile/FOXY i386
Architecture: i386
Machine: i386
>Description:
The tcom(4) driver probes and attaches com ports, but opening a tty
device causes fatal page fault in supervisor mode. It looks like the changes
in http://mail-index.netbsd.org/source-changes/2008/03/14/msg003488.html
were made for com(4) devices, but not for the devices that attached com(4)
as a slave device (i.e., multiport serial cards).
>How-To-Repeat:
Boot -current kernel on a machine that has a multiport serial card
that uses the tcom(4) driver, then watch it crash when the boot process
runs "ttyflags -a". Or boot single-user, and do "stty -f /dev/tty02"
(assuming tty02 is one of the tcom ports).
>Fix:
I'm not sure if this is correct, but it seems to work for me. I
think something similar will need to be done to the other multiport serial
cards in dev/isa that have the same basic structure. On further inspection,
it looks like ast(4) was already fixed (PR kern/38776); however, boca, ioat,
moxa, and rtfps look like they're still broken. My patch does things
differently than the way ast was fixed; perhaps the ast way is better;
I don't actually know anything about the kernel config mechanism.
Index: tcom.c
===================================================================
RCS file: /cvsroot/src/sys/dev/isa/tcom.c,v
retrieving revision 1.16
diff -u -r1.16 tcom.c
--- tcom.c 28 Apr 2008 20:23:52 -0000 1.16
+++ tcom.c 28 Sep 2008 01:37:59 -0000
@@ -91,28 +91,27 @@
#define STATUS_SIZE 8 /* 8 bytes reserved for irq status */
struct tcom_softc {
- struct device sc_dev;
+ device_t sc_dev;
void *sc_ih;
bus_space_tag_t sc_iot;
int sc_iobase;
int sc_alive; /* mask of slave units attached */
- void *sc_slaves[NSLAVES]; /* com device unit numbers */
+ device_t sc_slaves[NSLAVES]; /* com slave devices */
bus_space_handle_t sc_slaveioh[NSLAVES];
bus_space_handle_t sc_statusioh;
};
-int tcomprobe(struct device *, struct cfdata *, void *);
-void tcomattach(struct device *, struct device *, void *);
+int tcomprobe(struct device *, cfdata_t, void *);
+void tcomattach(struct device *, device_t, void *);
int tcomintr(void *);
-CFATTACH_DECL(tcom, sizeof(struct tcom_softc),
+CFATTACH_DECL_NEW(tcom, sizeof(struct tcom_softc),
tcomprobe, tcomattach, NULL, NULL);
int
-tcomprobe(struct device *parent, struct cfdata *self,
- void *aux)
+tcomprobe(struct device *parent, cfdata_t self, void *aux)
{
struct isa_attach_args *ia = aux;
bus_space_tag_t iot = ia->ia_iot;
@@ -181,9 +180,9 @@
}
void
-tcomattach(struct device *parent, struct device *self, void *aux)
+tcomattach(struct device *parent, device_t self, void *aux)
{
- struct tcom_softc *sc = (void *)self;
+ struct tcom_softc *sc = device_private(self);
struct isa_attach_args *ia = aux;
struct commulti_attach_args ca;
bus_space_tag_t iot = ia->ia_iot;
@@ -191,6 +190,7 @@
printf("\n");
+ sc->sc_dev = self;
sc->sc_iot = ia->ia_iot;
sc->sc_iobase = ia->ia_io[0].ir_addr;
@@ -199,14 +199,14 @@
if (!com_is_console(iot, iobase, &sc->sc_slaveioh[i]) &&
bus_space_map(iot, iobase, COM_NPORTS, 0,
&sc->sc_slaveioh[i])) {
- aprint_error_dev(&sc->sc_dev, "can't map i/o space for slave %d\n", i);
+ aprint_error_dev(sc->sc_dev, "can't map i/o space for slave %d\n", i);
return;
}
}
if (bus_space_map(iot, sc->sc_iobase + STATUS_OFFSET, STATUS_SIZE, 0,
&sc->sc_statusioh)) {
- aprint_error_dev(&sc->sc_dev, "can't map status space\n");
+ aprint_error_dev(sc->sc_dev, "can't map status space\n");
return;
}
@@ -242,7 +242,7 @@
for (;;) {
#define TRY(n) \
if (bits & (1 << (n))) \
- comintr(sc->sc_slaves[n]);
+ comintr(device_private(sc->sc_slaves[n]));
TRY(0);
TRY(1);
TRY(2);
>Release-Note:
>Audit-Trail:
From: Dave Huang <khym@azeotrope.org>
To: gnats-bugs@netbsd.org
Cc:
Subject: Re: kern/39642: tcom(4) driver is broken
Date: Sat, 27 Sep 2008 21:45:11 -0500
Here's a patch that stores the com_softc pointers in sc_slaves[], as ast(4)
does.
Index: tcom.c
===================================================================
RCS file: /cvsroot/src/sys/dev/isa/tcom.c,v
retrieving revision 1.16
diff -u -r1.16 tcom.c
--- tcom.c 28 Apr 2008 20:23:52 -0000 1.16
+++ tcom.c 28 Sep 2008 02:42:12 -0000
@@ -91,28 +91,27 @@
#define STATUS_SIZE 8 /* 8 bytes reserved for irq status */
struct tcom_softc {
- struct device sc_dev;
+ device_t sc_dev;
void *sc_ih;
bus_space_tag_t sc_iot;
int sc_iobase;
int sc_alive; /* mask of slave units attached */
- void *sc_slaves[NSLAVES]; /* com device unit numbers */
+ void *sc_slaves[NSLAVES]; /* com device softc pointers */
bus_space_handle_t sc_slaveioh[NSLAVES];
bus_space_handle_t sc_statusioh;
};
-int tcomprobe(struct device *, struct cfdata *, void *);
-void tcomattach(struct device *, struct device *, void *);
+int tcomprobe(struct device *, cfdata_t, void *);
+void tcomattach(struct device *, device_t, void *);
int tcomintr(void *);
-CFATTACH_DECL(tcom, sizeof(struct tcom_softc),
+CFATTACH_DECL_NEW(tcom, sizeof(struct tcom_softc),
tcomprobe, tcomattach, NULL, NULL);
int
-tcomprobe(struct device *parent, struct cfdata *self,
- void *aux)
+tcomprobe(struct device *parent, cfdata_t self, void *aux)
{
struct isa_attach_args *ia = aux;
bus_space_tag_t iot = ia->ia_iot;
@@ -181,16 +180,18 @@
}
void
-tcomattach(struct device *parent, struct device *self, void *aux)
+tcomattach(struct device *parent, device_t self, void *aux)
{
- struct tcom_softc *sc = (void *)self;
+ struct tcom_softc *sc = device_private(self);
struct isa_attach_args *ia = aux;
struct commulti_attach_args ca;
bus_space_tag_t iot = ia->ia_iot;
int i, iobase;
+ device_t slave;
printf("\n");
+ sc->sc_dev = self;
sc->sc_iot = ia->ia_iot;
sc->sc_iobase = ia->ia_io[0].ir_addr;
@@ -199,14 +200,14 @@
if (!com_is_console(iot, iobase, &sc->sc_slaveioh[i]) &&
bus_space_map(iot, iobase, COM_NPORTS, 0,
&sc->sc_slaveioh[i])) {
- aprint_error_dev(&sc->sc_dev, "can't map i/o space for slave %d\n", i);
+ aprint_error_dev(sc->sc_dev, "can't map i/o space for slave %d\n", i);
return;
}
}
if (bus_space_map(iot, sc->sc_iobase + STATUS_OFFSET, STATUS_SIZE, 0,
&sc->sc_statusioh)) {
- aprint_error_dev(&sc->sc_dev, "can't map status space\n");
+ aprint_error_dev(sc->sc_dev, "can't map status space\n");
return;
}
@@ -217,9 +218,11 @@
ca.ca_iobase = sc->sc_iobase + i * COM_NPORTS;
ca.ca_noien = 0;
- sc->sc_slaves[i] = config_found(self, &ca, commultiprint);
- if (sc->sc_slaves[i] != NULL)
+ slave = config_found(self, &ca, commultiprint);
+ if (slave != NULL) {
sc->sc_alive |= 1 << i;
+ sc->sc_slaves[i] = device_private(slave);
+ }
}
sc->sc_ih = isa_intr_establish(ia->ia_ic, ia->ia_irq[0].ir_irq,
From: Martin Husemann <martin@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc:
Subject: PR/39642 CVS commit: src/sys/dev/isa
Date: Sun, 28 Sep 2008 15:39:40 +0000 (UTC)
Module Name: src
Committed By: martin
Date: Sun Sep 28 15:39:40 UTC 2008
Modified Files:
src/sys/dev/isa: tcom.c
Log Message:
Apply patch from Dave Huang in PR kern/39642: catch up on com(4)'s
device_t/softc split.
To generate a diff of this commit:
cvs rdiff -r1.16 -r1.17 src/sys/dev/isa/tcom.c
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
State-Changed-From-To: open->closed
State-Changed-By: martin@NetBSD.org
State-Changed-When: Sun, 28 Sep 2008 15:41:08 +0000
State-Changed-Why:
second patch applied - thanks!
From: David Huang <khym@azeotrope.org>
To: gnats-bugs@NetBSD.org
Cc: kern-bug-people@netbsd.org,
netbsd-bugs@netbsd.org,
gnats-admin@netbsd.org,
martin@NetBSD.org
Subject: Re: kern/39642 (tcom(4) driver is broken)
Date: Sun, 28 Sep 2008 14:45:35 -0500
On Sep 28, 2008, at 10:41 AM, martin@netbsd.org wrote:
> Synopsis: tcom(4) driver is broken
>
> State-Changed-From-To: open->closed
> State-Changed-By: martin@NetBSD.org
> State-Changed-When: Sun, 28 Sep 2008 15:41:08 +0000
> State-Changed-Why:
> second patch applied - thanks!
Thanks :) What about the other drivers I mentioned that are similarly
broken? I could put together a patch if you're interested, although I
have no way of testing... I'm pretty sure that they need the same sort
of change though.
>Unformatted:
(Contact us)
$NetBSD: query-full-pr,v 1.39 2013/11/01 18:47:49 spz Exp $
$NetBSD: gnats_config.sh,v 1.8 2006/05/07 09:23:38 tsutsui Exp $
Copyright © 1994-2007
The NetBSD Foundation, Inc. ALL RIGHTS RESERVED.